1080 trouble shooting, possible memory issue?

[Korone609]

Hello all,

I finally got my first cad/gaming put together for college, running two used 1080's in sli. Everything worked fine for about a month. Now I am all of a sudden having horizontal line artifacts, black screen flashes, and occasional crashes. As well games will stutter and flip back to the desktop or frame rate plummet just to recover or CTD. When the system crashes gracefully enough for windows to give an error message I get, "Video Scheduler Internal Error." Both cards are running with nzxt krakens and AIO water coolers attached and seem to be hooked up properly. Issues usually occur in video playback and on startup. Here are my trouble shooting steps:

• DDU and install latest drivers
• DDU and roughly 5 year old drivers
• Use a drive with a clean windows install
  • worked fine for a bit. probably until it finished auto installing drivers
• test each card individually
  • same result as running the cards in sli
• using msi afterburner to underclock memory and core
  • this seems to give little to no impact or actually make the problem worse
• using msi afterburner to overclock memory and underclock core
  • this seems to actually give a bit of an improvement to stability, which confuses the heck out of me.
• testing different motherboard slots
• repasting one of the cards
• windows memory diagnostics
  • no memory issues reported
• heaven
  • surprisingly stable, with a few fps drops
• furmark
  • stable and consistent until it CTDs
• Running a basic blender project
  • Usually results in a CTD within 5 minutes
• monitoring temperatures using task manager
  • no temperature spikes and cards stay below 45 degrees
• Monitoring usage with task manager
  • Often a usage spike at CTDs, do not get to see on crashes

 

Additional ideas I am scared to test or haven't gotten to yet

• using msi afterburner to undervolt
• flashing card bios
  • I know this can go badly, especially if I am flashing on an unstable system
• more thoroughly cleaning the cards
• heat gun method
• testing a friends card in the system
• testing my cards in a friends system
• updating motherboard bios
• borrowing a different power supply
• Testing if Linux gets the same issues

 

Other system specs

• Ryzen 7 3800X with stock cooler
• 32GB of G skill memory
• Rog B550-E Gaming motherboard
• Thermaltake 850 Watt power supply

Any help would be much appreciated!

 

Update:

setting "performance enhancer" in bios from auto to default seems to have almost entirely fixed it. Occasionally after long hours of use in gpu heavy software such as Krita and blender I will see it pop back in again. Restarting so far has fixed these few flair ups. After doing this I have been able to overclock my memory by roughly 560MHz without being able to recreate the issue so I doubt it is a memory problem. If anybody has any ideas to how I can remove the few flair ups and get everything going smoothly it would be much appreciated. I would like to be able to mine to help pay my tuition but I don't want to accidentally kill my system if it is a memory issue.

Edited September 26, 2021 by Korone609
Further testing

[NRSV]

Hello

I'm no expert but see if these solutions might help you out

 

• Removing the SLI connecting the two cards
• Letting the motherboard run without with one GPU (that could be one of your 1080s)

[Korone609]

8 hours ago, NRSV said:

Hello

I'm no expert but see if these solutions might help you out

 

• Removing the SLI connecting the two cards
• Letting the motherboard run without with one GPU (that could be one of your 1080s)

Thank you for responding,

that was one of my first steps. I have tested both cards individually and sometimes I will think it has made a difference. And then 20 minutes later it's whacko again. Both cards seem to display the same issue to roughly the same degree.